Association of GSTM1 Null Variant With Anthracycline-Related Cardiomyopathy After Childhood Cancer–A Children’s Oncology Group ALTE03N1 Report

Abstract

BACKGROUND:

Anthracycline-related cardiomyopathy is a leading cause of late morbidity in childhood cancer survivors. Glutathione S-transferases (GSTs) are a class of phase II detoxification enzymes that facilitate the elimination of anthracyclines. As free-radical scavengers, GSTs could play a role in oxidative damage-induced cardiomyopathy. Associations between the GSTμ1 (GSTM1) null genotype and iron-overload-related cardiomyopathy have been reported in patients with thalassemia.

METHODS:

The authors sought to identify an association between the GSTM1 null genotype and anthracycline-related cardiomyopathy in childhood cancer survivors and to corroborate the association by examining GSTM1 gene expression in peripheral blood and human-induced pluripotent stem cell cardiomyocytes (hiPSC-CMs) from survivors with and without cardiomyopathy. GSTM1 gene deletion was examined by polymerase chain reaction in 75 survivors who had clinically validated cardiomyopathy (cases) and in 92 matched survivors without cardiomyopathy (controls). Conditional logistic regression analysis adjusting for sex, age at cancer diagnosis, chest radiation, and anthracycline dose was used to assess the association between genotype and cardiomyopathy. Proprietary bead array technology and quantitative real-time polymerase chain reaction were used to measure GSTM1 expression levels in samples from 20 cases and 20 matched controls. hiPSC-CMs from childhood cancer survivors (3 with cardiomyopathy, 3 without cardiomyopathy) also were examined for GSTM1 gene expression levels.

RESULTS:

A significant association was observed between the risk of cardiomyopathy and the GSTM1 null genotype (odds ratio, 2.7; 95% CI, 1.3-5.9; P = .007). There was significant downregulation of GSTM1 expression in cases compared with controls (average relative expression, 0.67 ± 0.57 vs 1.33 ±1.33, respectively; P = .049). hiPSC-CMs from patients who had cardiomyopathy revealed reduced GSTM1 expression (P = .007).

CONCLUSIONS:

The current findings could facilitate the identification of childhood cancer survivors who are at risk for anthracycline-related cardiomyopathy.

INTRODUCTION

Over 85% of children who are diagnosed with cancer today are expected to survive for 5 or more years.¹ Anthracycline chemotherapy (used to treat >50% of children with cancer) has contributed to the steady improvements in survival observed over the past several decades. However, cardiomyopathy progressing to congestive heart failure is an unfortunate complication of anthracyclines.^2,3 Indeed, childhood cancer survivors are at a 5-fold to 15-fold higher risk of anthracycline-related cardiotoxicity compared with age-matched controls without anthracycline exposure.^4–9 The outcome is poor, such that 5-year survival rates are <50% after a diagnosis of congestive heart failure.^10,11 Cardiomyopathy risk increases with anthracycline dose.^9,12,13 However, there is considerable interpatient variability in cardiomyopathy risk at any anthracyclines dose, suggesting that genetic factors possibly modify the association between anthracyclines and cardiomyopathy risk.^5,9,14–32

Anthracycline metabolism likely plays a significant role in the individual susceptibility to anthracycline-related cardiomyopathy. Indeed, high-risk genetic variants in the CBR3 gene result in the generation of larger amounts of cardiotoxic alcohol metabolites of anthracyclines³³ and have been implicated in increasing the risk of cardiomyopathy after moderate exposure to anthracyclines.⁹ Glutathione S-transferases (GSTs) are a class of phase II detoxification enzymes that catalyze the conjugation of reduced glutathione to toxic agents (including anthracyclines) and eliminate them from the body.³⁴ GSTs also act as scavengers of free radicals and could play a role in oxidative damage-induced cardiomyopathy—an important pathogenetic mechanism in anthracycline-related cardiotoxicity. Patients with thalassemia major are at risk for cardiomyopathy because of iron overload, and significant associations between the GSTμ1 (GSTM1) null genotype and left ventricular dysfunction have been reported in this population.^35,36 Homozygous deletions (null genotype) of the GSTM1 gene (chromosome 1p13.3) result in a complete absence of corresponding enzymatic activity.^37–39 However, an evaluation of the role of the GSTM1 null genotype in the development of anthracycline-related cardiomyopathy has not been undertaken. We used peripheral blood from anthracycline-exposed childhood cancer survivors who had cardiomyopathy (cases) matched to those without cardiomyopathy (controls) to genotype the GSTM1 gene and examine GSTM1 gene expression levels in peripheral blood and in human-induced pluripotent stem cell cardiomyocytes (hiPSC-CMs).

MATERIALS AND METHODS

Study Design and Population

Study participants were drawn from a Children’s Oncology Group (COG) study (ALTE03N1; Key Adverse Events After Childhood Cancer; clinicaltrials.gov identifier NCT00082745), NCT00082745; principal investigator, Smita Bhatia). COG member institutions (see Supporting Table 1) enrolled patients after obtaining approval from local institutional review boards. Written informed consent/assent was obtained from patients, parents, or legal guardians. A matched case-control study design was used. Cases consisted of anthracycline-exposed childhood cancer survivors who developed cardiomyopathy. For each case, from 1 to 3 anthracycline-exposed childhood cancer survivors with no signs or symptoms of cardiomyopathy were randomly selected from childhood cancer survivors registered with the COG and matched on primary cancer diagnosis, year of diagnosis (±5 years), and race/ethnicity. The selected controls were required to have had a longer duration of cardiomyopathy-free follow-up compared with the time from cancer diagnosis to cardiomyopathy for the corresponding matched case. Participants provided peripheral blood samples for DNA and RNA.

Phenotype Assessment

Anthracycline-exposed participants had normal cardiac function before anthracycline exposure. Cases fulfilled American Heart Association criteria for cardiac compromise⁴⁰ by presenting with symptoms (dyspnea, orthopnea, fatigue, etc) and/or signs (edema, hepatomegaly, rales, etc) of cardiac decompensation. None of the participants selected as controls had any symptoms or signs of cardiac compromise. Echocardiographic data were used as source validation for all cases and, when available, for controls.

Therapeutic Exposures

Lifetime anthracycline exposure was calculated by multiplying cumulative dose (mg/m²) of individual anthracyclines received by a factor that reflected the drug’s cardiotoxic potential⁴¹ and then summing the results. Chest radiation that included the heart was categorized as a yes/no variable.

DNA and RNA Isolation

Peripheral blood for DNA (n = 167) and RNA (n = 40) was collected directly into K2EDTA tubes and PAXgene Blood RNA tubes, respectively. DNA and RNA were isolated using the Gentra Puregene Blood Kit and the PAXgene Blood RNA kit (both from Qiagen Inc), respectively. The concentration was measured on Nanodrop ND-1000 Spectrophotometers (Thermo Fisher Scientific). DNA integrity was analyzed by using the Quant-iT PicoGreen dsDNA Assay Kit (Thermo Fisher Scientific). RNA quality was checked using a Bioanalyzer Nanochip (Agilent Technologies), and samples with RNA integrity numbers >7 were used for microarray analysis.

GSTM1 Gene Deletion

GSTM1 deletion was examined by multiplex polymerase chain reaction (PCR), as previously reported, using the B2M gene as control (for details, see Supporting Methods and Supporting Fig. 1).⁴²

Gene Expression Analysis

Microarray

Illumina HumanHT-12 v4.0 Expression BeadChips (Illumina, Inc) were used for mRNA transcription profiling (for details, see Supporting Methods).

Quantitative real-time PCR validation of GSTM1 gene expression

Expression of the GSTM1 gene also was measured using quantitative real-time PCR (for details, see Supporting Methods).

RNA sequencing of peripheral blood

Libraries were prepared using the TruSeq RNA Sample Preparation Kit (Illumina, Inc) according to the proto cols recommended by the manufacturer. Each library was paired-end sequenced (2 × 50 bp) by using the TruSeq SBS Kit v4-HS, on a HiSeq2000 platform.

RNA sequencing in cardiomyocytes

Day-30 hiPSC-CMs from 6 anthracycline-exposed childhood cancer survivors (3 with anthracycline-related cardiotoxicity and 3 without cardiotoxicity) were used.⁴³ RNA was extracted using TRI Reagent and the Direct-zol RNA microprep kit (both from Zymo), including on-column DNase digestion to remove genomic DNA. Samples were quantified using an Agilent 2100 Bioanalyzer and passed quality control. Gene expression levels and exon use were estimated using the featureCounts function in the Subread software⁴⁴. Differential gene expression analysis was performed using the DEseq2 package⁴⁵ and R (v3.3.3; R Foundation for Statistical Computing). Details are provided in the Supporting Methods.

Statistical Analysis

Analyses were conducted using GenomeStudio (Illumina, Inc.), R (http://www.r-project.org/, accessed July 2, 2019), SAS (SAS Institute Inc), and PLINK (http://zzz.bwh.harvard.edu/plink/, accessed July 2, 2019).. We summarized the characteristics of the survivor population according to case-control status. Continuous and categorical variables were examined using independent-sample t tests or Wilcoxon rank-sum tests and Fisher exact or chi-square tests, respectively.

Statistical analysis of genotype data

Multivariable conditional logistic regression analysis (adjusting for anthracycline dose, chest radiation, age at diagnosis of primary cancer, and sex) was used to determine the association between GSTM1 variants and cardiomyopathy. Odds ratios were calculated if the anthracycline dose was either dichotomized (<250 mg/m² and ≥250 mg/m²) or treated as a continuous variable. Analyses also were conducted for stratified cumulative anthracycline exposure (<250 mg/m₂ or ≥250 mg/m²).

Statistical analysis of gene expression data

To determine differential expression of the GSTM1 gene, we used paired t tests on log₂-transformed expression values.

RESULTS

Genotyping of Matched Case-Control Sets for GSTM1 Null Variant and Association With Cardiomyopathy Risk

DNA was available for 167 anthracycline-exposed childhood cancer survivors (75 cases and 92 matched controls). Demographic and clinical characteristics of the study participants are summarized in Table 1. The median age at primary cancer diagnosis for cases and controls was 7.8 and 9.6 years, respectively. The cumulative anthracycline dose was significantly higher among the cases (median, 300 vs 255 mg/m²; P = .02). Although a higher proportion of cases received chest radiation compared with controls (34.7% vs 26.1%), the difference did not reach statistical significance (P = .4). Controls were followed for a significantly longer time (median, 12 years) compared with cases (6 years; P < .0001). All cases were symptomatic; ie, they fulfilled American Heart Association criteria for cardiac compromise. The mean ejection fraction was 39.4%, and the mean fractional shortening was 22.2%. None of the controls had any clinical signs or symptoms of cardiac compromise. Echocardiographic indices available for 61% of the controls showed a mean ejection fraction of 65.9% and a mean fractional shortening of 36.7%. The echocardiographic validation of cases and controls are detailed in Supporting Table 2.

TABLE 1.

Demographic and Clinical Characteristics by Case-Control Status for Glutathione S-Transferase μ1 Genotyping

Variable	Cases, N = 75	Controls, N = 92	Pa
Age at primary cancer diagnosis, y
Mean ± SD	8.3 ±5.47	9.4 ±6.35	.2
Median [range]	7.8 [3.8-11.5]	9.6 [3.3-14.8]
Sex
Female	43 (57.3)	40 (43.5)	.07
Cumulative anthracycline exposure, mg/m2
Mean ± SDb	306.9 ± 111.6	264.4 ± 125.1	.02c
Median [range]	300 [230.3-375]	255 [150-368]
Race: No. (%)
Asian	4 (5.3)	6 (6.5)	.9
Black	2 (2.7)	4 (4.4)
Hispanic	4 (5.3)	6 (6.5)
White	65 (86.7)	76 (82.6)
Chest radiation
Yes: No. (%)	26 (34.7)	24 (26.1)	.4
Dose: Mean ± SD, cGy	1044.1 ± 1640.5	811.9 ± 1792.7
Primary diagnosis:
No. (%)
ALL	15 (20.0)	18 (19.7)
AML	5 (6.7)	6 (6.5)
Bone	20 (26.7)	20 (21.7)	.9
HD	8 (10.7)	10 (10.9)
Kidney	4 (5.3)	4 (4.3)
Neuroblastoma	6 (8.0)	9 (9.8)
NHL	8 (1 0.7)	14 (15.2)
Sarcoma	9 (12.0)	11 (11.9)
Time from diagnosis to cardiac event (cases) or time to enrollment (controls), y
Mean ± SD	7.42 ± 6.5	12.96 ± 7.1	<.0001c
Median [range]	6.0 [1.3-11.6]	12 [7.4-17.2]
Echocardiographic characteristics, %d
Ejection fraction	39.4	65.9
Fractional shortening	22.2	36.7

Abbreviations: ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; cGy, centigrays; HD, Hodgkin disease; NHL, non-Hodgkin lymphoma.

^aP values were estimated using either chi-square tests and t tests for categorical variables or the Wilcoxon test for continuous variables.

^bDoses were not available for 1 case and 1 control.

^cThis P value indicates a significant difference.

^dEchocardiographic indices were available for 100% of the cases and 61% of the controls; for the remaining controls, physician notes endorsed the absence of clinical signs/symptoms of cardiac compromise according to American Heart Association criteria.

There was overrepresentation of the GSTM1 null genotype among cases versus controls (60.0% vs 38.04%; P = .005) (see Supporting Table 3). Results of the multi variable conditional logistic regression analysis in Table 2 summarizes the association between the GSTM1 null genotype and the risk of cardiomyopathy. After adjusting for cumulative anthracycline exposure dichotomized as low-dose (<250 mg/m²) and high-dose (≥250 mg/m²), chest radiation (yes or no), age at diagnosis of primary cancer, and sex, the odds that patients who had the GSTM1 null genotype would develop cardiomyopathy were 2.7 times greater (95% CI, 1.3-5.9; P = .007) compared with patients who had the GSTM1 positive genotype. We also examined the association after including anthracycline dose as a continuous variable and found that patients with the GSTM1 null genotype had a 2.52-fold higher odds of developing cardiomyopathy (95% CI, 1.2-5.2; P = .01) (see Supporting Table 4) after adjusting for the variables listed above. Finally, conducting separate analyses for those exposed to high-dose anthracyclines and low-dose anthracyclines, the magnitude of GSTM1 cardiomyopathy was generally similar (see Supporting Table 5).

TABLE 2.

Risk Factors Associated With Anthracycline-Related Cardiomyopathy

Risk Factor	OR (95% CI)	Pa
Age at primary cancer diagnosis, y
Per y increase in age	0.95 (0.88-1.01)	.1
Anthracycline dose, mg/m2b
<250	1.00
≥250	2.5 (0.99-6.4)	.05
Chest radiation
No	1.00
Yes	1.5 (0.6-3.6)	.4
Sex
Male	1.00
Female	1.5 (0.7-2.9)	.3
Genotype
GSTM1 positive	1.00
GSTM1 null	2.7 (1.3-5.9)	.007c

Abbreviations: GSTM1, glutathione S-transferase μ1; OR, odds ratio.

^aP values were estimated from a conditional logistic regression model that included age at cancer diagnosis, sex, chest radiation and anthracycline exposure (categorical).

^bDoses were not available for 1 case and 1 control.

^cThis P value indicates a significant difference.

Constitutive Gene Expression Analysis of Matched Case-Control Sets for GSTM1 and Association With Cardiomyopathy Risk

Gene expression analysis using peripheral blood

This case-control set included 40 non-Hispanic white anthracycline-exposed childhood cancer survivors (20 cases and 20 matched controls) with high-quality RNA samples. Of these, 14 matched case-control pairs had also participated in the genotyping analysis. As shown in Table 3, the median age at primary cancer diagnosis for the cases and controls was 5.5 and 6.4 years, respectively (P = .8). Cases received a higher cumulative anthracycline exposure (median dose, 295 vs 217.5 mg/m²; P = .002). The median time between cancer diagnosis and cardiomyopathy was 11.4 years; controls were followed for a significantly longer period (median, 15.9 years; P = .02).

TABLE 3.

Demographic and Clinical Characteristics by Case-Control Status for Gene Expression Analysis

Variable	Cases, N = 20	Controls, N = 20	Pa
Age at primary cancer diagnosis, y
Mean ± SD	6.6 ± 5.4	7.5 ±5.8	.6
Median [range]	5.5 [1.9-9.2]	6.4 [1.4-13.2]	.8
Sex: No. (%)
Female	10 (50.0)	8 (40.0)	.5
Cumulative anthracycline exposure, mg/m2
Mean ± SD	304.6 ± 116.9	198.1 ±81.9	.002c
Median [range]	295 [240.7-350.0]	217.5 [121.3-266.3]	.002c
Chest radiation
Yes: No. (%)	7 (35.0)	7 (35.0)	1.0
Dose: Mean ± SD, cGy	797.5 ± 1280.7	725.5 ± 1218.3	0.9
Primary diagnosis:
No. (%)
ALL	5 (25.0)	5 (25.0)	1.0b
HL or NHL	9 (45.0)	9 (45.0)
Other	6 (30.0)	6 (30.0)
Time from cancer diagnosis, yd
Mean ± SD	12.5 ± 5.9	16.8 ± 5.7	.02c
Median [range]	11.4 [8.8-18.0]	15.9 [12.2-21.2]	.02c

Abbreviations: ALL, acute lymphoblastic leukemia; cGy, centigrays; HL, Hodgkin lymphoma; NHL, non-Hodgkin lymphoma.

^aP values were estimated using either chi-square tests and t tests for categorical variables or the Wilcoxon test for continuous variables.

^bThis P value was determined with the Fisher exact test.

^cThis P value indicates a significant difference.

^dThe time in years from cancer diagnosis to cardiac event is indicated for cases, and the time in years to enrollment is indicated for controls.

The analysis of all probes from the Illumina HT-12v4 expression array for the GSTM1 gene revealed 1 probe (ILMN_2391861) with signal intensities above those of the housekeeping genes. Figure 1 compares the average intensities for GSTM1 on the array in the 20 matched case-control pairs. GSTM1 showed significant downregulation in 12 of 20 cases compared with controls. Array-based differential expression for GSTM1 was confirmed by real-time quantitative PCR (see Supporting Fig. 2), with significant downregulation in cases versus controls (0.67 ± 0.57 vs 1.33 ± 1.33; P = .049). We also performed RNA sequencing (RNA-Seq) on 7 of the 20 paired samples. RNA-Seq reads ranged from 0 to 4 for null genotypes and from 62 to 180 for positive genotypes; genotypes were derived using the RNA-Seq reads with high confidence.

FIGURE 1.

The expression profile of the glutathione S-transferase μ1 (GSTM1) gene is shown on the Illumina Microarray (probe ILMN_2391861) in anthracycline-related cardiomyopathy. Matched cases are indicated in gray, and controls are indicated in pink.

Gene expression analysis using cardiomyocytes

hiPSC-CMs from patients who had anthracycline-related cardiotoxicity showed significantly reduced expression of GSTM1 (P = .007) (Fig. 2).

FIGURE 2.

Glutathione S-transferase μ1 (GSTM1) expression was determined by RNA sequencing in human-induced pluripotent stem cell cardiomyocytes derived from 6 childhood cancer survivors, including 3 anthracycline-exposed patients with no cardiomyopathy (NONTOX) and 3 anthracycline-exposed patients with cardiomyopathy (TOX).

DISCUSSION

In this study, we observed that the GSTM1 null variant was associated with anthracycline-related cardiomyopathy in anthracycline-exposed childhood cancer survivors. We extended these findings by demonstrating downregulated GSTM1 gene expression in the peripheral blood as well as reduced GSTM1 expression in hiPSC-CMs generated from childhood cancer survivors who had anthracycline-related cardiomyopathy. Taken together, these findings provide a biologically plausible association between the GSTM1 null variant and anthracycline-related cardiomyopathy.

GSTs are a class of phase II detoxification enzymes that catalyze the conjugation of reduced glutathione to toxic agents, including anthracyclines by making them water soluble and eliminating them from the body.³⁴ GSTs also act as scavengers of free radicals and could play a role in oxidative damage-induced cardiomyopathy. Homozygous deletions and the null genotype of the GSTM1 gene (chromosome 1p13.3) result in the complete absence of corresponding enzymatic activity.^37–39 There is significant variability in the GSTM1 null genotype by race/ethnicity and geography; eg, prevalence of the GSTM1 null genotype is approximately 5% in Europeans, 70% to 79% in East Asians, 57% in South Asians, 49% in Africans, and 69% in admixed American populations^46–48.

Previous investigations that sought to examine the as sociation with the GSTM1 null genotype were unsuccessful, in part because of the modest sample size of cases with cardiomyopathy^27,30–32 or because of challenges associated with analyzing copy number variation using genome-wide arrays.⁴⁹ However, significant associations between the GSTM1 null genotype and cardiomyopathy have been reported in patients with thalassemia.^35,36 Although some studies have examined the functional relevance of genetic variants,^{9,15,22,26,28,29} the functional relevance often remains unknown, in part because the identified single-nucleotide variations (formerly single-nucleotide polymorphisms) are located in intronic regions where they may play a role in regulating distant loci. Differential constitutive gene expression possibly could help discern the pathogenesis of anthracycline-related cardiomyopathy. Differential gene expression ideally should be measured in the affected tissue. However, procurement of cardiac tissue in anthracycline-exposed individuals is logistically difficult and not without risk, hence the dependence on surrogate tissues, such as peripheral blood. In addition, the establishment of methodologies to convert human-induced pluripotent stem cells (hiPSCs) to cardiomyocytes (hiP-SC-CMs) has enabled human cardiomyocytes to be mass produced in vitro for cardiovascular disease modeling^50.51.

This study needs to be considered within the context of its limitations. Members of the GSTμ class of enzymes share 75% to 99% sequence identity, with maximum homology between GSTM1 and GSTM2. Expression of GSTM1 (and GSTM2) was confirmed separately by real-time PCR, and null alleles for only GSTM1 were confirmed by genotyping; a null allele for GSTM2 is not reported. Although a successful replication of this association would have been ideal, the strong biologic plausibility and the demonstration of reduced expression in peripheral blood and in patient-derived cardiomyocytes strengthens the biologic basis of this association.

In summary, to our knowledge, we are the first to report an association between the GSTM1 null genotype and anthracycline-related cardiomyopathy. This finding, along with the previous reports of other genetic variants,⁶ could be considered in a risk-prediction model to facilitate the identification of childhood cancer survivors who are at increased risk of anthracycline-related cardiomyopathy, such that targeted interventions could be instituted.